Artico verso l'abisso... eppure lo dicevamo che...

[elz]

A proposito di impatti della perdita di ghiacci sulla circolazione atmosferica, c'è un'importante lavoro in pubblicazione che mostra quanto sia incerta la risposta remota alla perdita di ghiaccio artico e quanto sia piccola nei modelli in raffronto alla normale variabilità interna, alcuni dei punti fondamentali:

Observational studies suggest links between autumn sea-ice loss and circulation patterns in the following winter (Francis et al., 2009; Overland and Wang, 2010; Wu and Zhang, 2010: Strong et al., 2010; Jaiser et al., 2012), but the statistical significance of these linkages has been questioned (Hopsch et al., 2012), causality is unclear and the mechanisms are
poorly understood. In model simulations, the spatial pattern, strength, statistical significance and
timing of the circulation response to sea-ice loss differs considerably between studies, and can be
hard to disentangle from atmospheric internal variability (AIV).

Under doubled forcing, there is a weak cooling
response over mid-latitude Eurasia in DJF, but this is only significant over a limited area surrounding the Caspian Sea (Figure 2i).

In DJF, significant large-scale SLP decreases are found over the Arctic Ocean, Hudson Bay and eastern Canada in the UM (Figure 5h). Isolated regions of significant SLP reductions are also
identified over the Sea of Okhotsk and central North America. SLP is increased over Europe, but
this feature is not statistically significant. In the CAM, SLP decreases significantly over Hudson
Bay, Greenland and the Atlantic-side of the Arctic Ocean (Figure 5g).

Nmin for the SLP response is as low as 10 in the UM over regions of maximum ice loss, especially in
the double-perturbation case, but Nmin values this low are only found in very limited geographical
regions (Figure 5d-f; j-l). Generally, approximately 30-50 ensemble members are required to detect
a significant SLP response, and upwards of 50 members are required to detect a significant response in remote regions. It is notable that even with 100 ensemble members in the UM, very few mid-
latitude regions show a significant SLP response in the single-perturbation experiment. Further,
despite larger mid-latitude responses in CAM, an ensemble size of 60 is insufficient for these
achieve statistical significance. This implies that the remote SLP response to recent Arctic sea-ice
loss is considerably smaller than AIV.

The larger ensembles presented here do not support a shift towards to negative phase of the NAO in response to observed sea-ice loss. Instead, in CAM the response projects onto the positive
NAO phase and in the UM the response is not NAO-like. Thus, the wintertime circulation responses
(and their interactions with the large-scale modes of atmospheric variability) are not robust across
simulations, even those using the same models.

On this basis, we argue that an ensemble
size of around 50 members is desirable. This is considerably larger than the typical ensemble size
used in past studies of the atmospheric response to observed Arctic sea-ice loss (e.g., 5 in Ghatak et
al. (2012) and Orsolini et al. (2012), 5/8 in Screen et al. (2013), 10 in Strey et al. (2010), 15 in
Porter et al. (2012), 20 in Liu et al. (2012)).

e anche tutto il resto vale decisamente la pena leggerlo:
http://www.cgd.ucar.edu/cas/cdeser/D...tm_impacts.pdf